Religious markers reduce perceived trustworthiness in a Muslim-majority country

Significance

Recent studies suggest religiosity may signal benevolent intentions, thereby enhancing perceived trustworthiness. Using a large-scale survey experiment in Turkey, a predominantly Muslim society where religion is politically salient, we examine whether visible religious badges such as headscarves and beards function as signals of trustworthiness. Contrary to expectations, we find that religious markers often reduce perceived trustworthiness, even among highly religious respondents. We show that religious badges operate as bundled social signals that evoke more than religiosity and vary along gendered lines. These multivalent associations shape how individuals interpret religious markers in everyday life. In societies where public religious expression is contested, understanding how such markers shape social evaluations is key to identifying the boundaries of trust.

Abstract

First impressions often hinge on visible cues, leading people to infer the trustworthiness of strangers from their appearance and dress. While religiosity is generally associated with greater perceived trustworthiness, little is known about how visible religious markers, such as clothing and grooming styles, shape trust judgments in Muslim-majority societies, where such signals are widespread and easily recognizable. We examine this question using a large-scale vignette experiment embedded in a nationally representative face-to-face survey in Turkey. By experimentally varying profile characteristics, we identify the causal effect of a Muslim headscarf or beard on perceived trustworthiness. Contrary to expectations, individuals displaying these religious badges are generally viewed as less trustworthy than otherwise identical profiles, even by highly religious respondents. Mediation analyses show that these effects arise because religious badges signal multiple attributes, not only religiosity but also political orientation and physical attractiveness—and that these inferred meanings differ by gender: women’s headscarves and men’s beards prompt distinct patterns of inference. These findings show that visible religious markers shape trust judgments through multivalent social inferences rather than perceived piety alone. The results highlight how observers draw on broader social and political associations when interpreting religious markers, providing evidence on how public expressions of religion affect everyday trust judgments.

We frequently encounter situations that require us to place trust in others, often strangers. Trust underpins many forms of social and economic exchange: We must trust that an electrician will install the wiring correctly and safely in our new house or that a taxi driver will take the most direct route to our destination. Even in situations without economic exchange, we rely on trust, whether asking for directions or leaving belongings unattended in public places. Thus, social trust, the belief that others will uphold the trust placed in them, is a critical component of social exchange and a fundamental building block of our social world (1–4).

Often it is difficult to ascertain whether one’s interaction partner can be trusted, especially when they are a stranger. If we have no information about potential partners’ past behaviors in similar situations, we must rely on heuristic shortcuts to decide whom to trust (1). Trust decisions are often influenced by easily accessible information about a person that can be determined at first glance, such as their gender, age, race, or looks (5, 6).

One such characteristic that is widely believed to influence expectations of trustworthiness is religiosity (7, 8). There are several theoretical reasons to expect that religious people may be perceived to be more trustworthy than their less religious peers. First, many religious people adhere to the belief that their actions are being constantly watched by a supernatural entity and, as a result, they may fear divine retribution for dishonesty (9, 10). Second, religious people may have more opportunities to engage in prosocial behavior due to the involvement of religious communities in charitable work (11). These activities not only signal moral commitment but also provide tangible experiences in prosocial behavior and emphasize a greater sense of connectedness and interdependence (12). Finally, religious communities may attract inherently more prosocial individuals who seek this environment precisely because they wish to engage in these communities (13). By voluntarily associating with religious groups, these individuals reinforce the broader societal perception that religiosity correlates with trustworthiness. Together, these factors suggest that religiosity, whether due to actual behavioral differences or socially constructed expectations, plays a pivotal role in shaping perceptions of trustworthiness.

As a Turkish proverb goes, you can never tell who has money or faith. That is, a person’s true religious belief and degree of religiosity are not visible characteristics. However, many individuals wear religious badges or physical markers of religious group membership, which signal affiliation with a specific religious community or tradition (14). While intrinsic religious motivations often underlie this practice, they may also be worn for instrumental reasons. For instance, individuals may wear such badges to signal piety to their religious in-group, conveying that they can be trusted to adhere to shared religious customs (15, 16). Further, wearing a religious badge may act as a “commitment device,” reminding the wearer of the need to resist the temptation to stray from religious norms (15, 17). In some contexts, individuals may wear religious badges not only out of deep belief but also to gain benefits from policies that favor the devout (18), or to secure social advantages. Thus, while wearing a religious badge may be a costly signal of religiosity, as it may be stigmatized by out-group members (19), it is not always interpreted by observers as a sign of religiosity alone. Rather, the public expression of religion often functions as bundled social signals—evoking not only religiosity but also political orientation, lifestyle, and cultural values. As such, religious badges capture not just perceived piety, but a broader set of multivalent social meanings that observers interpret in real-world settings.

In this paper, we examine the effects of religious badges on perceptions of trustworthiness in Turkey. Although religiosity is often assumed to enhance perceived trustworthiness, it remains unclear whether visible badges reliably convey this signal or whether their effect varies with observer characteristics, such as their level of religiosity. Turkey is a strategic research site as a Muslim-majority country with a long history of public secularism. Religious badges like headscarves and facial hair are widespread, yet unlike in more conservative countries (e.g., Iran, Afghanistan), veiling is not mandatory; nor is it prohibited, as in Tajikistan. Although veiling was banned in Turkish government buildings until 2013, women now freely choose whether or not to cover their hair in public spaces. While most Turkish women cover their hair, about one-third do not (18). This variation in religious practice makes the choice to wear a religious badge a more powerful signal compared to contexts where veiling is either restricted or universal. As such, Turkey offers a unique opportunity to test the effects of religious badges in a Muslim-majority context, while building on previous research from predominantly Christian and Hindu contexts (20–22).

Following studies that suggest a positive relationship between religiosity and perceived trustworthiness (23–27), we find that, contrary to expectations, individuals displaying outward religious symbols were often perceived as less trustworthy than those without such markers. Our results from a large-scale vignette experiment, fielded as part of a face-to-face survey of a nationally representative sample of Turkish adults, show that this effect was especially pronounced for badges seen as more conservative expressions of religion, such as the chador or distinctively Islamic beard. Perceptions also varied by respondents’ religiosity but only for female vignettes: Religious respondents viewed women wearing a moderate badge (i.e., the türban) as more trustworthy than those who did not cover their heads or wore a conservative badge (i.e., the chador). Male vignettes wearing religious badges, however, were generally perceived as less trustworthy regardless of observer religiosity. Mediation analyses reveal that religious badges signal not only religiosity but also political orientation and physical attractiveness and that observers interpret women’s headscarves and men’s religious beards differently.

These findings show that there is no positive universal link between visible religious badges and perceived trustworthiness. Instead, such badges function less as unambiguous signs of piety than as bundled social signals. Their meaning depends on a complex interplay of observer characteristics, differences in how religious badges function across genders, the specific type of badge, and the broader political and social context. This multivalence is central to our contribution: It reflects how individuals interpret religious cues in everyday interactions. Empirically, our results contribute to ongoing public debates about religion in public life by showing that visible religious markers do not consistently signal trustworthiness; in some contexts, they may heighten group boundaries rather than promote trust and social cohesion.

Setting and Research Design

We investigated judgments of trustworthiness as reactions to visible religious markers in a preregistered^* face-to-face survey experiment with 2,170 Turkish adults. We employed a full factorial design combining textual and visual information. Respondents each evaluated six vignettes presented in printed booklets, with responses manually recorded.

To manipulate religiosity, facial profiles of four women and four men were edited to create three versions of each face, corresponding to three levels of religiosity. We added visual markers of religiosity that are common in the Turkish context: head coverings for women and facial hair for men. For both genders, the nonreligious condition included no veil or beard. For women, the religious profile featured a headscarf (türban) that covers the hair, while the devout profile displayed a black chador, which covers all but the face. For men, the religious profile included a full beard with a mustache, and the devout profile showed a full beard with a trimmed mustache, a style observed among most conservative Muslims. Example photographs for each condition are shown in Fig. 1.

Fig. 1.

Experimental religiosity badge treatments for male and female vignette persons on sample pictures: Left, no religious badge; Center, religious badge (türban for women, full beard with mustache for men); Right, devout (chador for women, full beard with trimmed mustache for men). Treatments applied to sample photos from the Bogazici Face Database (28).

To assess whether our visual treatments successfully manipulated perceived religiosity, respondents rated each vignette person’s perceived religiosity on a 0 to 10 scale. As shown in Fig. 2, both veils and beards significantly increased perceived religiosity, with headscarves producing a stronger effect. Beards appear as a more ambiguous religious signal, but the manipulation successfully differentiated three levels of perceived religiosity for both female and male profiles.

Fig. 2.

Perceived religiosity (on a scale from 0 to 10) by vignette gender and religiosity badge. Women wearing headscarves, whether türban (religious) or chador (devout), are rated significantly more religious than women without veils. For men, profiles with a beard, religious or devout, are rated as more religious than nonbearded profiles. The profiles with the devout badge for both female and male profiles are perceived to be the most religious. $N=12,782$ ratings from 2,133 respondents.

The design controlled for social class and physical attractiveness, two potential confounders. Social status was signaled by one of six common occupations: three low-status (caregiver, receptionist, and sales clerk) and three high-status (doctor, engineer, and lawyer). Facial attractiveness was varied at two levels (low and high, see Materials and Methods), as it is known to influence trustworthiness perceptions (29, 30). Based on the orthogonal manipulation of vignette religiosity via badges, social class via occupation, and facial attractiveness, the estimated effects on trustworthiness are not confounded by social class or appearance.

For each vignette person, respondents were asked to rate their first impressions on scales from 0 to 10. We measured perceived trustworthiness using a widely adopted “lost wallet question” (2, 31, 32), which asks (translated from Turkish): “Imagine that you lose your wallet in a public place. If this person finds your wallet, what is the probability that they will return it to you with everything in it?” This measure improves upon abstract trust items by grounding respondents’ judgments in a concrete situation with clearly defined roles and outcomes. It specifies who is trusted, to do what, and under what circumstances (33–35), helping to ensure more consistent interpretation across respondents (2). This measure has been shown to positively correlate with standard survey and experimental measures of trust (36, 37).

While the scenario does not involve a deliberate act of entrusting (as in a strategic exchange or experimental trust game), it nevertheless captures a core dimension of generalized trust: The belief that others will act honestly even when they could benefit from dishonesty. It places the respondent in a position of vulnerability, whereby the vignette person possesses a valuable item and must decide whether to return it without external incentives. Although this vulnerability is accidental rather than strategic, it still reflects expectations about the person’s honesty and willingness to help a stranger. By focusing on expected behavior in a realistic situation, rather than directly assessing self-reported trustworthiness, this measure mitigates social desirability bias and more accurately captures the core concept of generalized trust (2).

Results

We first examine whether religious badges increase perceived trustworthiness, as suggested in prior research (20–22, 38). Multilevel regression models with random intercepts for respondents and neighborhoods reveal the opposite pattern for male vignettes: Bearded men are rated as significantly less trustworthy than men without beards (religious: $\beta =-0.68$, 95% CI [−0.86; −0.50], devout: $\beta =-1.28$, 95% CI [−1.47; −1.10]; Fig. 3, SI Appendix, Table S2). For women, effects vary by head covering. Those wearing the more common türban are rated about as trustworthy as unveiled women ($\beta =0.09$, 95% CI [−0.06; 0.24]), whereas women in the more conservative chador are rated significantly less trustworthy ($\beta =-0.94$, 95% CI [−1.16; −0.72]). Those patterns are consistent across male and female respondents (SI Appendix, Fig. S1). Marginal means (SI Appendix, Fig. S2) show that the effects of religious badges on perceived trustworthiness are at least as strong and often stronger than the effects of social class and facial attractiveness.

Fig. 3.

Left two panels: Perceptions of trustworthiness by vignette gender and religiosity treatment (veil or beard), compared to nonreligious vignettes (reference category). N = 12,910 ratings from 2,163 respondents. Right two panels: Perceptions of trustworthiness by vignette gender and religiosity treatment by respondents’ self-reported religiosity, compared to ratings of nonreligious vignettes by less religious respondents (reference category). N = 12,801 from 2,141 respondents. Error bars indicate 95% CIs.

Next, we examine the influence of social class, signaled by the vignette person’s occupation. Manipulating social class experimentally allows us to rule out confounding between religiosity and socioeconomic status in Turkey (39, 40). As shown in SI Appendix, Fig. S3 and Table S4, higher social class significantly increases trustworthiness ratings ($\beta =0.71$, 95% CI [0.62, 0.80]), with consistent effects across levels of religiosity and vignette gender (see SI Appendix, Table S4, Models 3 to 4). Notably, higher-class devout female profiles are perceived as similarly trustworthy as lower-class nonreligious or religious female profiles, highlighting the relevance of class for trustworthiness judgments alongside religious badges. Including vignette social class in the models does not affect the estimated effects of religious badges (SI Appendix, Table S2, Models 2 and 6; SI Appendix, Table S3, Model 2), as expected, given that these two dimensions were orthogonal by design in the factorial survey.

We also test whether social class moderates the effects of religious badges on trustworthiness evaluations, based on the idea that higher-class individuals may face less scrutiny than do lower-class individuals when wearing religious markers. However, we find no evidence that social class plays such a moderating role (see SI Appendix, Table S3, Model 5).^† We further examine whether respondents’ own social class shapes their evaluations. Across multiple indicators, including self-reported social class, standardized monthly income, interviewer-rated perceived social class, and education, we find no moderation of respondents’ social class on the effects of religious badges on trustworthiness judgments (SI Appendix, Figs. S4–S7).

Since religious badges may also influence perceptions of facial attractiveness, a known predictor of trustworthiness, we included attractiveness in our design, manipulated independently at two levels (high vs. low). As expected, given this orthogonal design, including attractiveness in the models (SI Appendix, Table S2, Models 3 and 7; SI Appendix, Table S3, Model 3) does not alter the main effects of religiosity on trustworthiness. Highly attractive vignette profiles are rated as more trustworthy than less attractive ones, and although attractiveness can be associated with perceived social class (41), our analyses show that the effects of facial attractiveness and social class are additive rather than interactive (SI Appendix, Table S5). Moderation tests (SI Appendix, Fig. S8 and Table S6) indicate that higher attractiveness consistently increases trustworthiness for male vignettes, while for female vignettes in religious and devout conditions, attractiveness does not significantly modify the effects of religious badges on trustworthiness.

Building on our main findings, we next examine whether the effect of religious badges on perceived trustworthiness varies by respondent religiosity. We hypothesized that more religious respondents might react more strongly to religious badges, and thus would rate more religious vignette persons as more trustworthy. To test this, we split our sample into respondents above and below the mean value of self-reported religiosity (3.60 on a scale from 1 to 5). Fig. 3 (Right two panels) reports the coefficients separately for both respondents below and above mean religiosity.^‡

Consistent with the overall pattern, individuals displaying religious badges are generally perceived as less trustworthy than those without. However, observer religiosity moderates these perceptions. More religious respondents rate female vignettes with religious or devout badges as significantly more trustworthy than do less religious respondents (SI Appendix, Table S7, Model 2). They also view women wearing the türban as more trustworthy than those without religious markers. A similar pattern emerges for male vignettes in the devout condition, where more religious respondents rate them as more trustworthy. However, for male vignettes in the religious condition, no significant difference is found between respondents below and above mean religiosity. While the general pattern holds that all respondents, regardless of their religiosity, tend to view nonreligious vignettes as more trustworthy, with the exception of female vignettes with the türban, religious respondents appear to find religious and devout vignettes more trustworthy than do less religious respondents.

To assess the robustness of our findings, we test alternative operationalizations of respondent religiosity. Using behavioral markers, namely, whether respondents reported wearing a veil ($n=620$, or 56% of women in the sample wear a veil) or a beard ($n=673$, or 66% of men have a beard), we find results that are substantively similar to those based on self-reported religiosity (SI Appendix, Table S8). Similar patterns hold when using interviewer ratings of respondents’ religiosity (SI Appendix, Table S9). Finally, these findings remain robust under alternative model specifications including sample weights and interviewer fixed effects (SI Appendix, Table S10).

Exploratory Analyses.

We also examine whether respondents’ political attitudes shape perceptions of religious badges. Despite Turkey’s long history of laïcité, the current government led by the conservative Justice and Development Party (AK Parti, or AKP) has expanded the role of Islam in public life, and support for the AKP is associated with religiosity (18). Supporters of the ruling AKP and its coalition partners may thus judge religious badges more positively than do voters of opposition parties. We find that respondents who indicated that they voted for the AKP in the last general election ($n=595$, or 40% of those who voted) rated women wearing religious markers as significantly more trustworthy than did voters of other parties ($\beta =0.55$, 95% CI [0.25, 0.85] vs. $\beta =-0.20$, 95% CI [−0.44, 0.05]), as shown in SI Appendix, Fig. S12 and Table S11. They also rated women in the devout condition as similarly trustworthy as unveiled women. For male profiles, religious badges reduced perceived trustworthiness even among AKP voters. Although trustworthiness ratings in the religious condition did not differ significantly by political affiliation, AKP voters rated devout men as more trustworthy than opposition voters. These findings suggest that political orientation moderates the effect of religious badges on trustworthiness perceptions: AKP voters expect greater trustworthiness from veiled women and from bearded men in the devout condition.

To better understand our unexpected findings, we conducted causal mediation analyses (42) to examine what information religious badges convey. The results indicate that respondents’ perceptions of the vignette person’s religiosity, political attitudes, and facial attractiveness all mediate the relationship between religious badges and trustworthiness expectations, indicating both direct and indirect causal pathways.

As shown in SI Appendix, Table S12, the direct effect of religious badges on trustworthiness remains negative when controlling for perceived religiosity. However, the indirect effect through perceived religiosity is positive and statistically significant for both female ($\beta =0.82$, 95% CI [0.74, 0.90]) and male ($\beta =0.22$, 95% CI [0.17, 0.27]) profiles ($P<0.001$ in the religious condition). The contrast between the strong negative direct effects and the significant positive indirect effects through perceived religiosity suggests that although religious badges generally reduce trustworthiness ratings, they also convey religiosity, which enhances perceived trustworthiness.

We also examine whether perceived support for the opposition coalition mediates the effect between religious badges and perceived trustworthiness (SI Appendix, Table S13). Results show a small but statistically significant negative indirect effect: Vignettes in the religious condition are seen as less likely to support the opposition, which reduces perceived trustworthiness. This suggests that political perceptions partially mediate the trustworthiness penalty.

Finally, SI Appendix, Table S14 shows that religious badges reduce perceived attractiveness, which in turn lowers trustworthiness evaluations. In this model, the direct effect of the türban becomes positive and the direct effect of both types of beards is near zero, suggesting that religious badges could have neutral or even positive effects on trustworthiness absent their negative effect on perceived attractiveness.

Taken together, these results indicate that religious badges convey complex, multivalent cues: They increase perceived religiosity, which enhances trustworthiness, but they also signal political orientation and reduce perceived attractiveness that together shape trustworthiness judgments.

Discussion

In everyday life, individuals must often decide whom to trust with limited information. This forces them to rely on visual cues and heuristics to make judgments. In a large-scale nationwide face-to-face survey experiment conducted in Turkey, we provide evidence for the relationship between religious badges and perceived trustworthiness. Contrary to our initial expectations, religious badges, such as the chador on women and religious beards on men, are not interpreted to signal trustworthiness. In fact, individuals with these markers are perceived to be less trustworthy than their counterparts without such badges. Interestingly, women wearing a türban are rated as similarly trustworthy as unveiled women, suggesting that the type of religious badge matters in shaping trust perceptions.

While it would seem at first glance that our results challenge a wealth of prior research that reports a positive association between religiosity and perceived trustworthiness (20, 23–27), results from our mediation analysis in fact support this finding. To the extent that religious badges convey information about religiosity, they are also associated with higher trustworthiness. However, we find that this information is outweighed by other information about the wearer’s perceived political leanings and perceived physical attractiveness that negatively affect trustworthiness ratings. Thus, our results do not align with previous research that finds a positive association between religious badges and trustworthiness in other contexts (21, 22, 38). In Turkey, we find a trust premium only for türban-wearing women among religious respondents. Our study finds a null to negative association between religious badges and perceived trustworthiness even within a religious in-group. This result may reflect unique features of the religious badges examined: Both veils and beards can obscure facial features, thereby reducing perceived attractiveness, which itself influences trustworthiness judgments.

In addition, respondents’ own religious and political beliefs play a role in shaping interpersonal perceptions of trustworthiness. Among highly religious respondents and supporters of the pro-Islamic and conservative AKP, a trust premium exists for veiled women but is limited to those wearing a türban; women in a chador do not benefit. These findings suggest that religious badges are not consistently interpreted as trust-enhancing signals, even within religious in-groups. Perhaps most strikingly, no trust premium exists for bearded men. In fact, bearded men are perceived as less trustworthy even by highly religious respondents and AKP supporters.

One plausible explanation for these findings lies in Turkey’s sociopolitical history, where religious badges carry layered meanings. The rise of the openly pro-Islamic AKP, which overturned previous bans on veiling, has likely polarized views of veiled women, associating the headscarf with both religious piety and political identity (43, 44). What was once primarily a sign of religious devotion has increasingly taken on political connotations, a shift supported by our findings that AKP supporters have the most positive judgments of veiled women. This pattern seems consistent with the tenets of social identity theory (45), which would suggest that individuals place more trust in members of a shared in-group, perhaps by shaping expectations of reciprocity (3, 46). However, our results challenge a simplistic group-based interpretation: Religious badges do not enhance perceived trustworthiness among all religious in-group members. Instead, trust premiums for religious badges appear contingent on an overlap of both religious and political identities, and on the specific badge worn. This in-group premium only applies to women wearing the türban and not to all religious badges, thus suggesting that women’s religious symbols are more politicized than men’s beards in public life.

A further explanation is that Turkey is becoming more secular, marked by declining private religiosity in recent years (44, 47) despite substantial government investment in religion (48) alongside intensified political polarization around religion (43). In our sample, while more than 96% of respondents identify as Muslim, more than 40% report praying never or only occasionally (e.g., once a month or on religious holidays). This suggests that many identify with Islam culturally rather than through regular religious practice, which may foster skepticism or lower trust toward those perceived as highly devout. In this context, public religious symbols, such as veils, may also serve as political signals or strategic displays of conformity in an environment, where outward religiosity can yield political or social benefits (18). This helps explain why religious badges do not consistently increase perceived trustworthiness and may even backfire, especially in less religious communities, as reflected in the variation of the treatment effect we find across districts with differing levels of religiosity (SI Appendix, Fig. S11).

Signaling theory (49–51) provides another explanation for interpreting religious markers. It suggests that religious badges may no longer function as reliable signals of trustworthiness when they are no longer costly and can therefore be produced as easily by mimics. This distinction is especially relevant when some markers, such as facial hair, require sustained investment, while others, like garments, can be adopted or removed depending on expectations that may vary across situations and places (i.e., rural vs. urban contexts). This may explain the differences between beards and veiling we find in this study. However, the informational value of such markers depends not only on their intrinsic costliness but also on the social meaning attached to them. In contexts where both veils and beards are widespread and rarely stigmatized, even markers that require sustained investment may convey little information about moral commitment. Moreover, while garments could be easier to take off and put on than shaving and sporting a beard, having seen as abandoning a religious garment could be very costly in a religious community. Under such conditions, choosing not to display a religious badge may, in fact, be more stigmatized and hence more “costly,” thereby carrying greater signaling value. These patterns again point to the notion that the interpretation of badges in Turkey reflects the bundled social meanings observers attach to them, rather than their mere costliness.

Research on moral licensing (52–54) offers an additional perspective: Individuals signaling high moral standards may feel licensed to engage in morally questionable behaviors in mundane situations. Observers may interpret religious badges as moral compensation for antisocial behavior (such as keeping a lost wallet) rather than indicators of trustworthiness, particularly in a polarized sociopolitical context. While moral licensing offers a potential explanation for some of the observed patterns, such as why respondents penalize wearers of more visible devout attire, such as the chador, it does not fully account for all patterns observed.

We also uncover strong gendered dynamics. Specifically, veiling provides a more unambiguous signal of religiosity for women, whereas beards are a weaker, more ambiguous and less consistent religious marker for men (Fig. 2). While the türban increases perceived trustworthiness among religious respondents and AKP supporters, beards in any form are associated with a trust penalty even for religious respondents and AKP supporters. Beards may signal masculinity, dominance, or threat (55), which can reduce perceptions of trustworthiness. The plausibility of this badge, especially in the most devout form, may be further limited in professional contexts (e.g., doctors, lawyers), where such a beard is rare and may seem incongruent with occupational norms. Moreover, bearded profiles are rated as particularly less attractive, and when we control for perceived attractiveness, the negative effects of religious badges on trustworthiness attenuate, suggesting that perceived attractiveness mediates this badge-trust link (SI Appendix, Table S14).

Like all studies, ours has some limitations. Religious badges increase perceived religiosity, but because social status extends beyond someone’s occupation, it may still shape interpretations of these markers despite our experimental design explicitly varying the occupation of vignette profiles. Additionally, the religious badges we study convey complex meanings beyond religiosity and political orientation. Future research should consider alternative signals, including gender-neutral markers that do not affect perceived attractiveness, to better isolate gendered perceptions of religious signals. In the Turkish context, the particular styles we used, the black chador and the mustache-less religious beard, may carry additional associations with political radicalism, sectarian identity, or foreignness, potentially activating outgroup perceptions among both secular and religious respondents. Moreover, the devout markers may even be less common among higher-status professional groups. This is an important consideration, as social class is conveyed through information on occupation in our vignette design, potentially limiting the ecological validity of the devout condition in this context. We recommend future studies explore different veiling styles, colors, and additional nationality cues to reduce ambiguity about group membership and capture more nuanced variation in perceived religiosity.

Given that Turkey is not necessarily representative of other Muslim-majority contexts, our findings should be tested in other settings, where religious badges may carry greater signaling costs (56). Comparative studies across religious traditions and diverse political-cultural environments could clarify whether religious badges primarily signal trust to in-group members (22, 38) or whether their effects depend on the costliness of display, as signaling theory predicts (49–51).

In sum, our study challenges the notion of a universal trust premium for religious badges. Just as the effects of religiosity on prosocial behavior are diminished in environments with high levels of social enforcement (57), the relationship between religious badges and trust perceptions also depends on political, cultural, and social conditions of the context. Future research should examine more closely how political climate, social norms, and individual experiences interact to shape the meaning and reception of religious markers. While prior research often treats religiosity as an unambiguous signal of prosociality, our findings show that, under certain sociopolitical circumstances, such signals can be perceived negatively. Religious badges should therefore be understood not as inherently trust-enhancing, but as complex, contextually embedded symbols whose meanings, shaped by multiple dimensions of identity, influence trustworthiness judgments.

Materials and Methods

Respondents were randomly assigned to one of 24 survey booklets, each of which included six vignette persons (three female, three male), with one in each religiosity condition. The vignettes were printed in a randomized order, and interviewers presented the vignettes in the order in which they were printed. Each booklet ensured a balanced distribution of facial attractiveness, with three highly attractive and three less attractive profiles, and prevented respondents from seeing the same face across multiple treatments.

Each vignette included a half-A4-size color photograph alongside a brief description varying gender, religiosity, social class, and physical attractiveness. The vignette text read: “The person whose photo you see is working as a [caregiver]. We will ask you to evaluate this person.”

In addition to the lost wallet question, respondents rated the vignette person’s attractiveness (how “beautiful” for women, or “handsome” for men), perceived religiosity, and political leanings. Finally, we collected demographic information on respondents’ own religiosity, political attitudes, and socioeconomic status.

Face Imagery.

We sourced face imagery from the Bogazici Face Database, a collection of standardized headshot-style photos of Turkish undergraduate students (28). This database includes ratings from 1,207 Turkish adults on various appearance-related factors, including attractiveness, masculinity, femininity, and trustworthiness. Based on these ratings, we selected 16 photos (8 male, 8 female) that provided the desired variation in terms of physical attractiveness. These photos were then edited to standardize their clothing: We removed all jewelry and replaced their clothing with white collared shirts, styled differently for males and females. This editing was intended to minimize any unintended social class signals. We also cropped out most of the subjects’ shirts to focus attention on the face. We then had our selection of edited photos rated again by a sample of 550 Turkish adults drawn from an online access panel, to ensure that these changes did not substantially affect how they are perceived with respect to attractiveness.

We selected 8 photos from an initial set of 16, ensuring two for each combination of gender (male/female) and physical attractiveness (low/high). These photos were edited then by a professional designer to apply the religious treatments. Each of the 8 faces appeared in three conditions (without religious markers, religious, and devout), resulting in a total of 24 photos for the vignette experiment.

Ethical Approval.

The design of this study was approved by the Ethics Committee of the University of Wuppertal on October 2, 2023.

Sampling and Fieldwork.

To ensure broad representation of the Turkish adult population, we used a stratified sampling strategy that randomly selected respondents within 119 neighborhoods in 30 provinces across the country, including at least one province within each of Turkey’s NUTS 1 regions. We also stratified by community size to ensure the inclusion of small villages (population less than 2,000), small urban areas, and large metropolitan areas. Additionally, we applied quotas for age and gender, surveying nine men and nine women in each neighborhood, three of each in the age categories 18 to 33, 34 to 50, and 51 and over.

Fieldwork was completed from June 28 to 30, 2024. Surveyors from the survey company Konda used a random walk strategy to select respondents, inviting respondents in randomly selected homes to participate. They continued to solicit responses until they filled the quotas in their neighborhood, achieving a response rate of 43%. All respondents were provided with a short summary of the survey and offered their consent to participate. Surveyors conducted 2,263 interviews but they removed 93 interviews for data quality concerns. We thus achieved a final sample of 2,170 responses. To verify data quality, we independently checked the representativeness of the sample on key variables against official statistics such as gender and age distributions and voting patterns. Summary statistics for the full sample are shown in SI Appendix, Table S1.

Statistical Analysis.

We first constructed a directed acyclic graph with DAGitty (58) to determine whether any variables could potentially confound the estimated effect of religious badges on perceived trustworthiness (SI Appendix, Fig. S13).

To account for our data structure, which consists of multiple vignette ratings nested within respondents and respondents nested within sampling neighborhoods, we fit multilevel regression models with random intercepts for respondents and neighborhoods. All SEs reported are robust to clustering at the neighborhood level. For missing data, we used case-wise deletion. Data analysis was conducted using Stata (version 18) (59). Mediation analyses were implemented using the mediate command in Stata, which fits causal mediation models and estimates direct, indirect, and total treatment effects. Robustness checks presented in SI Appendix, Table S10 were estimated using the REGHDFE package (60).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Replication files including the full questionnaire and all code and data used to produce the analyses presented in the manuscript and SI Appendix can be found on OSF: https://osf.io/buxwj/?view_only=e1d9bd4ba8a9413ba25bb468c0f8fa5d (61).